Open Books: Conceptsearch

The ConceptSearch / Open Books website allows users to quickly query the ECCO (Eighteenth Century Collections Online) corpus of over 200,000 novels split into millions of passages, drilling down into the dataset to find relevant passages. The technique allows users to construct searches from entire passages rather than just keywords, or recursively build up searches. The site also includes many filters and some experimental datavis of vectorized passages.

I was a lead developer on the project, provisioned the Elasticsearch and Elastic Beanstalk infrastructure, and worked on both Django and NLP features. Dr. Stephen Osadetz (PI), Christine Fensebrener-Eslao, and I presented on the project at DH 2018 Mexico City.